ICDL2008 Influence of ageing on conduction and breakdown in rape-seed and mineral oils C. Tran Duy 1 , A. Denat 1 , O. Lesaint 1 , N. Bonifaci 1 , Y. Bertrand 2 , W. Daoud 3 and M. Hassanzadeh 3 1 Grenoble Electrical Engineering Laboratory (G2E lab), Grenoble University and CNRS, 25 avenue des Martyrs, 38042 Grenoble (France). 2 EDF R&D, Les Renardières - Bat. 054, 77818 Moret sur Loing (France) 3 AREVA T&D, 1340 rue de Pinville, 34965 Montpellier (France) chau.tranduy@grenoble.cnrs.fr Abstract- The purpose of this work is to evaluate the chemical stability of a bio-degradable insulating oil composed by a mixture of esters based on rape-seed oil. The measurements are done in comparison with a commercial mineral oil. Both oils are subjected to accelerated ageing conditions at 120°C under air (open beaker method). Dielectric properties (dissipation factor, permittivity, conductivity, ac breakdown voltage) and chemical parameters (acidity, water content) are monitored during the ageing process. In the rape-seed based oil, ageing has a negligible influence on the breakdown voltage, whereas conductivity and acid content show a large increase. 1. INTRODUCTION Vegetable-based insulating oils are considered today for insulation of medium voltage systems. Compared to mineral oils, they show higher dielectric losses, higher conductivity, higher viscosity but a larger ability to dissolve water and a higher flash point [1-4]. However, as the high viscosity of such vegetable oils is detrimental in some applications, a modified vegetable oil (RS50) based on an oleic rape-seed oil has been developed [5]. Its kinematic viscosity (17mm 2 /s at 40°C) is close to values of mineral transformer oils, and it has a much higher flash point (>170°C). The purpose of this work is to compare the chemical stability of RS50 oil, a blend of esters based on rape-seed oil with a commercial mineral oil. Both oils were subjected to accelerated ageing conditions at 120°C under air (open beaker method) for a period extending over 50 days. During the test, dielectric properties such as dissipation factor, permittivity and conductivity have been monitored by a spectrometer over a large frequency range from 10 -2 Hz to 10 6 Hz. Soluble acid and water contents have also been measured. In addition, a study of ac breakdown voltage -measured according to the IEC 60156 standard [6]- has been carried out in order to verify the impact of ageing on dielectric strength. II. EXPERIMENTAL TECHNIQUES All ageing experiments were carried out in comparison with Univolt54, a commercial mineral oil with high oxidation stability and low acidity. The accelerated ageing test is the following. A one litre glass vessel is filled with 0.8l of liquid. In this way, the liquid sample has an exchange surface with air of 150 cm 2 . The vessel is then placed in an oven at fixed temperature (120±0.5 °C). The ageing process lasted up to 50 days. The vegetable and mineral oils used are free of antioxidant. The oil chemical stability is then estimated by measurement of dielectric properties, soluble acidity and ac breakdown voltage. Dielectric measurements are carried out by using a test cell made up of two stainless steel electrodes (coaxial cylinders) fitted into a faraday cage. The electrode gap is 3.5 mm and the capacitance in air is 22 pF. Measurements were carried out at 120°C. A broadband impedance spectrometer (Novocontrol) was used. In liquids, a typical frequency spectrum of the dissipation factor tg shows two branches [7]: one at low frequency f corresponding to conduction losses (relation 1) and the other at high frequency due to dipolar losses (relation 2). In RS50 oil, dielectric spectra have shown conduction losses for f<1kHz and polarization losses for f>10kHz [8]. s s f tg ε π σ ωε σ δ 2 = = (1) τω ε ε ω δ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ - ≅ = ∞ s b tg 1 (2) Here, is conductivity, s and • are static and optical permittivity respectively, pulsation, f frequency and is dipole relaxation time. The ac breakdown voltage was measured according to the IEC 60156 standard, except that the distance between electrodes was reduced to 1mm instead of 2.5mm. The breakdown test cell was included in a closed-loop filtering circuit (pore diameter of 10μm). The liquid was circulated in the filtering circuit before and after each breakdown measurement. Water content being an essential parameter in such measurements, it was systematically monitored using the Karl Fischer method. The water content of samples was adjusted over a wide range (10 to 110% of saturation) by using either a drying system